Software Engineering graduate (TAMK, May 2026) with a keen interest in the invisible layers of tech: cloud architecture, infrastructure security, networking, and embedded hardware. I like connecting the gap between intelligent software systems and raw, physical hardware — building backend systems, cloud infrastructure, and local AI pipelines that work in the real world, not just in demos.
Currently building out hands-on enterprise-grade topologies, deploying custom Linux homelabs, and automating systems from the ground up. Pursuing AZ-104 (Azure Administrator Associate).
Portfolio: https://tahrimwalid.github.io/ · Email: tahrimwalid@gmail.com · LinkedIn: walid-hassan-tahrim · Tampere, Finland
Languages & Scripting
Frontend & Infrastructure-as-Code
Systems, Cloud & DevOps
AI & Edge
Kratos — Self-hostable, self-growing AI Security Analysis System with a TUI
Evolved from Bachelor's Thesis
A terminal security assistant that investigates Linux machines in plain language. Ask a question and an agentic loop picks from 19 read-only tools (auth logs, open ports, processes, file integrity, YARA, CVE matching, privileged accounts), reached over SSH or a dial-out sub-agent for boxes behind NAT. A rule engine with no model in it (21 correlation rules, including sliding-window SSH burst detection) turns observations into findings, and 11 answer checks reject conclusions the evidence doesn't back: a claim about the wrong time window, a dismissed HIGH finding, a count that doesn't match. Works with any OpenAI-compatible model, hosted or local. /evolve writes a missing tool, tests it in a network-less sandbox, and keeps it only on your approval. Also: deterministic audits, schedules, phone alerts, and an MCP server. ~2,400 tests.
InGen — AI-Powered AWS Infrastructure Compiler
Live: in-gen-five.vercel.app
A visual AWS architecture designer (ReactFlow) that validates against 32 deterministic rules and compiles to production-ready Terraform HCL. Bidirectional sync between diagram and code via Monaco, 800ms debounced, with loop prevention. AI layer runs on self-hosted Qwen3-27B-FP8 via vLLM, scoring 10/10 on complex 12-resource serverless architectures.
Career AI — Finnish IT Job Market Analysis Pipeline
A 5-phase ETL pipeline scraping ~2,000 Finnish IT job postings via a reverse-engineered undocumented API, processed with a local LLM into 1,998 structured records (96.7% hydration, 100% extraction success). Mid-project, identified and documented a deduplication strategy that swung a key bias metric from 5% to 82.5% to 4.5% across three technically valid runs — prioritizing honest unmeasurability over false precision.
€5,000 funded by the TAMK Foundation · team project · private repo
Built the backend benchmarking layer: 40+ REST API endpoints across 9 functional categories (market benchmarks, org comparisons, rankings, gap analysis, trend tracking) designed retrieval-first so an LLM grounds every benchmarking claim in live structured data rather than general knowledge. Deployed as a systemd daemon behind Cloudflare Tunnel and Apache on Ubuntu.
(Code available on request)
Networking & Cloud Security Cisco IOS, Subnetting, Routing Protocols (OSPF, BGP) (active track) · Secure Remote Access, VPN Tunneling, Network Hardening
Systems & Infrastructure Linux (Ubuntu, Debian, Server Administration) · Docker, Containerized Microservices · Physical Server Architecture, Custom Homelab Deployment & Hardware Engineering
- Networking: building advanced virtual topologies in GNS3/EVE-NG — routing, switching, core protocols, CCNA track
- Homelab: re-engineering low-power SFF server nodes for multi-drive high-density storage via PCIe expansion
- DevOps: automating local deployments with Bash and Docker to replicate cloud-native environments locally
- Certification: pursuing AZ-104, redeploying Career AI onto Azure infrastructure as hands-on lab practice
Tampere, Finland — would love to work with cloud architecture, DevOps and low level LLM deployment at scale.